Microsoft Certified Azure AI Fundamentals · Free Practice Question Easy
Question 8
Among the following regression metrics, which one is most sensitive to large errors?
-
A
Coefficient of Determination (R²)
-
B
Root Mean Squared Error (RMSE)
-
C
Mean Absolute Error (MAE)
-
D
Mean Squared Error (MSE)
Reveal correct answer
Correct answer: D
Explanation
When evaluating regression models, one critical metric is how sensitive the error measure is to large deviations between the predicted and actual values. Mean Squared Error (MSE) is especially sensitive to large errors because it squares each individual error before averaging. This squaring operation means that even a few predictions with large errors can disproportionately increase the overall error metric, making MSE an excellent indicator when large errors are particularly undesirable.
By contrast, Mean Absolute Error (MAE) treats all errors linearly, which means it does not penalize larger errors as harshly. Similarly, the Coefficient of Determination (R²) measures the proportion of variance explained by the model and does not provide a direct quantification of error magnitudes. Root Mean Squared Error (RMSE) is derived by taking the square root of MSE, which brings the error units back to the original scale, but it still retains the sensitivity characteristic of squaring.
In scenarios where it is critical to identify and minimize large prediction errors—for instance, when even a single significant error could have major implications—MSE is the most appropriate metric. Its high sensitivity ensures that models with occasional large deviations are penalized more severely, encouraging improvements that reduce such extreme errors.
Further Reading and Documentation:
For additional insights into regression metrics and their properties, review the Evaluate Model component documentationlearn.microsoft.com, which outlines various evaluation metrics used in Azure Machine Learning.
More detailed discussions on regression evaluation metrics and their trade-offs can also be found in the Azure Machine Learning overview
Microsoft's Educational Content Around This Topic For The AI-900 Exam:
A.
R² measures the proportion of variance explained by the model and does not directly quantify the magnitude of errors. It is not designed to be sensitive to the size of individual errors.
B.
RMSE is the square root of MSE and is also sensitive to large errors. However, since it is on the same scale as the target variable, its sensitivity is somewhat normalized compared to MSE. MSE remains the most sensitive metric to large errors due to the squaring operation.
C.
MAE computes the average absolute difference between predicted and actual values. While it provides a clear measure of overall error, it does not disproportionately penalize larger errors as it treats all errors linearly.
D.
MSE squares each error before averaging, which means that larger errors have a much greater impact on the overall metric. This squaring makes MSE highly sensitive to outliers and large errors.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
